Una procedura di modellazione automatica degli edifici con dati LIDAR

3D building modelling is nowadays considered an important topic for all those application related to the landscape administration; lidar is one of the most used data acquisition techniques which is able to generate points cloud required for the generation of building and landscape digitai models. The need to process huge amount 01 data collected with lidar stimulated many researchers to develop modelling algorithms able to produce results in (semi)automated way. For 3D building modelling in particular, many methods for the segmentation of laser data and the generation of roof models have been proposed. The present work concerns the implementation of a procedure composed by different modules able to extract building roof models by using previously filtered lidar data. By supposing that building roofs can be defined with planes related to each other by adjacency relations, the roof digital model is realized by following the subsequent steps: buildings or building blocks identilication, segmentation of points for roof planes definition, extraction of exterior roof planes polylines, generation 01 the final 3D model. Once the segmenlation step has been performed with a neighbouring criteria, a region-growing based algorithm is used IO detect the rocl planes; outline boundaries 01 the rocf planes are obtained by applying a simplification algorithm named etshapes and two alternative procedures, RANSAC and Sleeve-Firting, to identify each line that will compose the regularized outline boundary of the roof plane. The main purpose of the implemented method is to generate 3D models by identifying each plane separately without making any a priori assumption for the roof shape.